Research Projects

It is proposed that a novel approach towards significantly improving ultrasonic imaging in the field of non-destructive testing and evaluation will be investigated. Ultrasonic Imaging is a key methodology in the field of NDT&E, where quality assurance during manufacturing, and in-service inspection, ensures reduced wastage, increased life, improved safety, thereby significantly reducing energy costs and material wastages. The novel approach involves the merger of two technologies viz. (a) the use of focused virtual sources based beam-forming inside the material that is optimally designed (AVASA), and (b) the use of DPAI based models that is used to compute the optimal beam-forming design focal-laws using AI engines that are pretrained. Prof. Krishnan Balasubramanian has pioneered both AVASA and DPAI methods and separately demonstrated it successfully in solving NDT&E problems, particularly in weld inspection. Traditional phased array-based use active focusing beam-forming methods to achieve real-time high-resolution images. The synthetic beam forming approach (FMC/TFM method) is fast replacing this active method, but has 4 short-comings (a) limits the ultrasonic energy that can be imparted (b) fixed configuration transducer (c) periodicity sidelobes, (d) slow performance. The AVASA technique proposed here overcomes all of the above 4 limitations. This novel AVASA method, developed by Prof. Krishnan Balasubramanian and his team, combines the active and the synthetic beam forming approach. The AVASA method employs high energy virtual sources that are created by actively focusing the ultrasonic beam at any location within the material, thereby overcoming the first shortcoming of the synthetic only beam forming, and can be employed to penetrate thick or attenuative materials. By designing an optimal array of focus locations, the second limitation is addressed, wherein the beam can be designed for every problem even by using linear array transducer. The third limitation is also eliminated by using virtual source array configurations that may be arbitrary or random. Finally, the fourth shortcoming is reduced, as reported in several publications from his group, by using reduced number of virtual sources, while achieving excellent Signal to Noise Ratio (SNR) in the images. DPAI uses AI Engines that are trained using a small set of reduced size numerical FEM models and permits significant time and resource reduction in computation of large models. Hence, in this proposed work, the deployment of DPAI based edge hardware for simulation and optimization of the AVASA based ultrasonic imaging will be explored and demonstrated. Both AVASA and DPAI have been patented and published and these two powerful techniques will be merged into a single tool that will provide a edge-optimized ultrasonic imaging technique that promises to revolutionize inspection of complex problems and applied in other fields such as radar, bio-medical and geological applications.